Nonparametric Spectral Analysis of Periodically Nonstationary Vibration Signals for Electrical Rotary Machines Testing

Abstract

The estimation of the instantaneous spectral density of periodically random processes (PCRP) and its Fourier coefficients in continuous and discrete cases is considered. The formulae for calculation of the mean square error dependencies on parameters of processing and signals are obtained. The aliasing effects are analyzed. The example of estimation of PCRP spectral function for vibration signal of defect rotary unit is given.